Bitcoin Price Analysis and Prediction Using DMA and LSTM

This paper delves into analyzing and predicting cryptocurrency prices, focusing on Bitcoin. It employs displaced moving average (DMA) and long short-term memory (LSTM) techniques. By leveraging these methods, we aim to provide valuable insights for those navigating the volatile world of cryptocurrency investments. After extracting it using Python within a Jupyter Notebook environment, a dataset sourced from Yahoo Finance is utilized. The analysis incorporates 50 DMA and 200 DMA to enhance understanding of bitcoin market dynamics and offers insights into the predictive capabilities of moving averages. Subsequently, the study applies LSTM, a recurrent neural network combining forget and output gates, to enhance memory and assess attributes critical for accurate price forecasting. The study systematically assesses model performance using important metrics like MSE (Mean Squared Error), R-squared (R2) factor, and the RMSE (Root Mean Squared Error) for understanding accuracy and reliability. For comparison, the traditional regression method is also analyzed for Bitcoin price forecasting, revealing the outperformance of LSTM over regression.